Real wage flexibility in the enlarged EU: evidence from a structural VAR.

Babecky, Jan ; Dybczak, Kamil

Membership in the monetary union imposes higher demands on factor market flexibility, since neither the exchange rate nor monetary policies can be used to deal with country-specific shocks. In this paper we assess the ability of the twelve new EU member states (NMS-12) to dampen the impact of shocks by means of macroeconomic wage flexibility. Following the structural VAR approach elaborated in Moore and Pentecost (2006), real wage flexibility is measured by the responsiveness of real wages to real (permanent) and nominal (temporary) shocks. The analysis of Moore and Pentecost (2006) is extended in three ways: by employing a new Eurostat labour cost data set covering 1996QI to 2007Q3, by using a large sample of 24 EU member countries, and by assessing the sensitivity of the results to the sample length. We find evidence of heterogeneous real wage adjustment across the new as well as the mature EU economies. Overall, the degree of real wage flexibility in the NMS- 12 lies within the bounds of the corresponding values for the Euro Area 'core' and 'peripheral' member countries.

Keywords: Wage flexibility; structural VAR

JEL Classifications: E24; C22; F02; J30; P20

I. Introduction

Membership in the monetary union imposes higher demands on factor market flexibility, since neither the exchange rate nor monetary policies can be used to deal with country-specific shocks. A number of studies stress the importance of higher labour market flexibility in the context of the EMU (e.g. Hallett et al., 2000; Obstfeld, 1997; Pissarides, 1997), of a currency board arrangement (e.g. Guide et al., 2000), or of a less rigid exchange rate peg such as the European Exchange Rate Mechanism (e.g. Kopits, 1999). Indeed, it is commonly argued that a fixed exchange rate regime eliminates one degree of freedom in absorbing macroeconomic shocks. Since independent exchange rate policy is no longer available under fixed exchange rate arrangements, adjustment through the labour market should be of a higher magnitude in countries with fixed exchange rates than with flexible ones. Joining the monetary union imposes further limits on aggregate demand management, as both autonomous monetary and exchange rate policies become unavailable. Therefore, there is a call for higher labour market flexibility in countries which are members of the monetary union or those which intend to join the monetary union.

The notion of labour market flexibility is of course a very broad one. In principle, the labour market can accommodate shocks via two main channels: either quantities (adjustment in workers and in working time), or prices (wages), or a combination of both. Due to limited and even declining mobility of workers within the new member states, and given the formal restrictions on the free movement of labour for new EU members, it is unlikely that migration can be considered an efficient tool for coping with adverse shocks. (1) Hence, more interest is focused on wage flexibility.

Hyclak and Johnes (1992), Boeri et al. (1998), and Blanchflower (2001) argue that wage flexibility is a key determinant of labour market flexibility. In particular, adjustment in prices might seem quicker and less costly than adjustment in quantities. The European Commission (2003, p. 155) stresses the importance of wage flexibility in the following paragraph:

"Obviously, wages as the price of labour have a key role to play in determining the overall balance of supply and demand on the labour market. Furthermore, the formation of economic and monetary union (EMU) is often taken to put further demands on the flexibility of wages to compensate for the lack of (national) instruments to deal with economic disturbances. If wages are too rigid, the necessary adjustment will come slowly and with considerable economic and social costs."

Wage flexibility can be expressed in nominal or real terms. Nominal wage flexibility is the responsiveness of nominal wages to changes in the price level or inflation. Real wage flexibility can, in turn, be defined as the responsiveness of real wages to various shocks (e.g. shocks to productivity, unemployment, past wages, etc.) Wage flexibility reflects different factors if measured using aggregate or micro data. Groshen and Schweitzer (1996) extensively discuss both macro- and microeconomic approaches to wage flexibility from the viewpoint of employees as well as employers. (2) In addition, institutional factors such as the minimum wage, the tax and benefit system and social protection system are often analysed when assessing labour market flexibility. Indeed, institutions can significantly affect the overall labour market flexibility (Dickens et al., 2006; Pentecost and Sessions, 2002; or Dessy, 2005). Nevertheless, in our study we focus on the flexibility of wages, leaving the institutional set-up aside.

From the macroeconomic point of view, "aggregate real wage flexibility determines the overall balance of supply and demand in the labour market and is a key substitute for the adjustment roles of the nominal exchange rate and an independent monetary policy" (HM Treasury, 2003, p. 2). Since the difference between real and nominal wage growth is given by inflation, real and nominal wage adjustment approach each other in a low inflation environment. This paper, therefore, focuses on real wage adjustments, and the analysis is conducted for the twelve new EU member states (NMS-12) and compared to the twelve mature Euro Area members (EA-12).

The measures of real wage flexibility, which are based on the responsiveness of real wages to shocks in real variables such as productivity, unemployment, etc., do not allow one to distinguish between the shocks themselves and the reactions to them, since both components are present in the macroeconomic time series. In this study we adopt the structural VAR approach used by Moore and Pentecost (2006), in order to asses the responsiveness of real wages to structural shocks. In particular, real wage flexibility is defined in relation to so-called real (permanent) and nominal (transitory) shocks. (3) Real wages are called flexible if the variance in real wages is mainly due to real shocks. On the contrary, if nominal shocks explain most of the variance in real wages, such a situation corresponds to rigid real wages. Thus, the degree of real wage flexibility is given by the percentage of the variance in real wages that can be attributed to real shocks.

Moore and Pentecost (2006) use wage flexibility to assess the suitability of the Czech Republic, Hungary, Poland and Slovakia for membership in the Euro Area with France and Italy considered to be benchmarks. (Although wage flexibility is important, it is obviously not a sufficient condition for a country to join the monetary union.) If real wages in, for example, Hungary are as responsive to real shocks as in, for example, Italy, then Hungary is said to be 'suitable' for EMU membership. Based on wage flexibility alone, the Czech Republic and Hungary are found to be good candidates for the EMU, while euro adoption is not advisable for Poland and Slovakia. However, France and Italy belong to the so-called core Euro Area countries; per capita incomes in France and Italy are higher than the Euro Area average.

What seems to be more relevant is, first, to compare the new EU member states with the Euro Area peripheral countries, such as Austria, Greece and Portugal. Second, it is important to assess how heterogeneous wage flexibility is across the Euro Area countries. We extend the analysis of Moore and Pentecost (2006) in three ways. First, we employ a new Eurostat labour cost data set covering 1996Q1-2007Q3. Secondly, we use a large sample of 24 EU member countries. Finally, we assess the sensitivity of the results to the sample length. We find evidence of heterogeneous real wage adjustment across the NMS-12 as well as the EA-12. Overall, the degree of real wage flexibility in the NMS-12 lies within the bounds of the corresponding values for the Euro Area 'core' and 'peripheral' member countries. Also, there is evidence of rising real wage flexibility in the NMS-12 group.

The paper is organised as follows. The following section describes the data set and gives the stylised evolution of labour costs in the EU-24. Section 3 provides details on the concept of real wage flexibility and describes the identification strategy. The empirical results are presented in Section 4 and the last section concludes.

2. Data description

Our sample includes twelve new EU member states (NMS-12), namely Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, Slovakia and Slovenia and twelve mature Euro Area member states (EA-12), namely Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal and Spain. (4)

[FIGURE 1 OMITTED]

In order to measure real wage flexibility, we need a variable characterising the development of labour costs both in nominal and real terms. For this purpose, we use the labour cost index provided quarterly by the Eurostat. In addition to wages and salaries (the indicator used in Moore and Pentecost, 2006), the labour cost index also includes employers' social security contributions plus taxes paid less subsidies received by the employer. The Eurostat index is available in both nominal and real terms, and the data have the advantage of being harmonised for a cross-country comparison. Nominal and real indices are seasonally adjusted and adjusted by working days, and normalised to 100 in 2000. The two variables that interest us, expressed in natural logarithms, are shown in figure 1. (For brevity, we will refer to the real and nominal labour cost indices as real and nominal wages.)

It is not surprising that growth in nominal wages in the twelve new EU member states is higher compared to the corresponding numbers for the twelve mature member states. A number of countries belonging to the NMS-12 group, in particular Bulgaria and Romania, experienced high inflation episodes during the 1990s. Real wages

also tend to grow faster in the NMS-12 group compared to their EA-12 counterpart. Economies with the highest rates of real wage growth are the Czech Republic, Estonia, Latvia and Lithuania.

To assess the time series properties of the data, we apply standard techniques: the augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests, and the Kwiatowski-Phillips-Schmidt-Shin (KPSS) stationarity test. (5) Overall, the series of log nominal and real wages can be characterised as integrated of order one (of order two in the case of Latvia and of order zero for Italy), although we acknowledge that eleven years of data might be too short a period for a robust reference.

3. Stylised representation of the aggregate labour market and the identification of shocks

In order to identify structural shocks from the observed fluctuations in nominal and real wages, Moore and Pentecost (2006) apply a bivariate structural vector autoregressive (SVAR) procedure. This identification strategy in turn is based upon a bi-variate SVAR decomposition proposed by Blanchard and Quah (1989), in the way that Bayoumi and Eichengreen (1996) apply this decomposition to extract real (supply) and nominal (demand) shocks from quarterly series of real output and prices. Such an approach is quite popular in studies of business cycle convergence for developed as well as emerging economies. (6) In our case, structural shocks are defined according to their short- and long-term effect on nominal and real wages. By definition, one type of shock (labelled as 'nominal') has only a transitory impact on the level of real wages, while another type of shock (labelled as 'real') might have a long-term impact on the level of real wages.

Notice that real shocks can affect real wages in either positive or negative directions. A positive effect can be associated, for example, with a rise in productivity, followed by a permanent increase in real wages and employment. This leads to an outward shift of the aggregate labour demand curve. A negative impact of the real shock on real wages can be interpreted as due to an increase in labour supply, followed by a decrease in real wages.

Although nominal shocks cannot have long-lasting effects on real wages, no restrictions are imposed on the short-run effects and their sign and magnitude depend on relative price/wage stickiness. If real wages WR= W/P decrease following a positive nominal shock, such a situation corresponds to sticky nominal wages W. Under a sticky price assumption, real wages increase m response to a positive nominal shock. Lastly, if nominal wages W and prices P move simultaneously, real wages do not change.

Economic theory proposes alternative explanations as to why markets do not clear immediately after an unexpected shock hits the economy. In particular, New Keynesians claim that rigidity of wages and prices is one of the most relevant causes of economic fluctuations, for example, the sticky wages and sticky prices assumption. (7) On the one hand, the sticky wages assumption imposes rigidity on the short-run adjustment of wages to demand shocks, thanks to implicit or explicit agreements in the labour market. On the other hand, the sticky price assumption imposes rigidity on the short-run price adjustment to demand shocks mainly due to menu cost. Although the two assumptions appear quite similar, their real economic implications are in sharp contrast. As discussed for example by Kandil (1996), the real wage can develop procyclically or countercyclically depending on the adjustment of nominal wages and prices. Under the assumption of sticky wages a temporary demand shock translates into higher prices and lower real wage rates, i.e. the real wages move countercyclically. In contrast, under sticky prices a positive demand shock tends to increase demand for labour and real wages. Thus, under sticky prices assumptions real wages and other real economic variables move procyclically.

A structural bivariate VAR decomposition makes it possible to identify real (permanent) and nominal (transitory) shocks from the observable movements of real and nominal wages. Formally, let us consider [wr.sub.t] and [w.sub.t], real and nominal wages expressed in logarithms. These variables are assumed to be first difference stationary. The following VAR representation can be estimated:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

where [e.sup.wr.sub.t] and [e.sup.w.sub.t] are white-noise disturbances, [b.sub.ijk] are coefficients, and K is the lag length chosen so that [e.sup.wr.sub.t] and [e.sup.w.sub.t] are serially uncorrelated. (8) Disturbances [e.sup.wr.sub.t] and [e.sup.w.sub.t] are not structural, they simply represent unexplained components in real and nominal wage growth movements. In order to recover structural disturbances, i.e. those having an economic interpretation of real and nominal shocks, the following two relationships are proposed:

[e.sup.wr.sub.t] = [c.sub.11][[epsilon].sup.N.sub.t] + [c.sub.12][[epsilon].sup.R.sub.t] (3)

[e.sup.w.sub.t] = [c.sub.12][[epsilon].sup.N.sub.t] + [c.sub.22] [[epsilon].sup.R.sub.t] (4)

where [[epsilon].sup.N.sub.t] and [[epsiolon].sup.R.sub.t] are nominal (transitory) and real (permanent) disturbances respectively. These equations state that the unexplainable components in the movements of real and nominal wage growth are linear combinations of structural shocks. In matrix form, [e.sub.t] = C[[epsilon].sub.t] . The vector of the structural disturbances et can be obtained by inverting matrix C: [[epsilon].sub.t] = [C.sup.-1][e.sub.t].

In order to recover the four coefficients of matrix C, four restrictions have to be imposed. Knowledge of the variance-covariance matrix of the estimated disturbances [[epsilon].sup.N.sub.t] and [[epsilon].sup.R.sub.t] is sufficient to specify three restrictions:

[c.sup.2.sub.11] + [c.sup.2.sub.12] = V ar ([e.sup.wr] (5)

[c.sup.2.sub.21] + [c.sup.2.sub.22] = V ar ([e.sup.w] (6)

[c.sub.11][c.sub.21] + [c.sub.12] [c.sub.22] = Cov ([e.sup.wr], [e.sup.w] (7)

These restrictions on the coefficients of matrix C are directly derived from (3) and (4) using normalisation conditions:

(i) the variance of nominal and real shocks is unity: Var([[epsilon].sup.N] = Var([[epsilon].sup.R] = 1

(ii) nominal and real shocks are orthogonal: Cov([[epsilon].sup.N], [[epsilon].sup.R]) = 0

The fourth restriction on the coefficients of matrix C is that nominal shocks [[epsilon].sup.N.sub.t] have no long-term impact on the level of real wages. To put this restriction into a mathematical form, one should substitute equations (3) and (4) into the VAR system given by (1) and (2), and then express variables [wr.sub.t] and [w.sub.t] as the sum of the contemporaneous and past realisations of structural disturbances [[epsilon].sup.N.sub.t] and [[epsilon].sup.R.sub.t]:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (8)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (9)

System (8)-(9) is an infinite moving-average representation of the VAR form (1)-(2). Coefficients [c.sub.ijk]--called impulse response functions--characterise the effect of structural disturbances on the left-hand-side variables after k periods ([c.sub.ijk] can be expressed in terms of the four coefficients of interest [c.sub.ij] and the estimated coefficients [b.sub.ij]). The restriction that the cumulative effect of nominal disturbances on real wage growth is zero, for all possible realisations of demand disturbances, means that

[[infinity].summation over (k=0) [c.sub.11k] = 0. (2)

This restriction also implies that nominal disturbances have no long-term impact on the level of real wages itself. Indeed, [C.sub.11k] represents the effect of the nominal disturbance [[epsilon].sup.N.sub.t-k] on [DELTA][wr.sub.t] = [wr.sub.t]-[wr.sub.t-1], real wage growth after k periods. Therefore, the sequence [c.sub.110], [c.sub.111], [c.sub.112],...,[c.sub.11k],[c.sub.11k] represents the effect of [[epsilon].sup.R.sub.t-k] on [wr.sub.t]-[wr.sub.t-1], [wr.sub.t+1]-[wr.sub.t], [wr.sub.t+2]-[wr.sub.t+1],..., [wr.sub.t+k-2], [wr.sub.t+k]-[wr.sub.t+k-1]. Hence, the cumulative restriction

[[infinity].summation over (k=0) [c.sub.11k] = 0.

states that the effect of [[epsilon].sup.N.sub.t] on [wr.sub.t+K]-[wr.sub.t-1] equals zero, i.e. that the level of real wage does not change in the long run: [wr.sub.t-1] = [wr.sub.t+K]. It can furthermore be shown that the restriction

[[infinity].summation over (k=0) [c.sub.11k] = 0.

translates into the parameters of interest [c.sub.ij] and the coefficients [b.sub.ijk] of the unrestricted VAR system (1)-(2) as:

[c.sub.11] (1- [K.summation over (k=0) [b.sub.22k]) + [c.sub.21] [K.summation over (k=0) [b.sub.12k] = 0. (10)

Restrictions (5), (6), (7) and (10) serve to identify the four coefficients [c.sub.ij] which, in turn, are used to recover the real and nominal disturbances from the VAR residuals by inverting matrix C: [[epsilon].sub.t] = [C.sup.-1] [e.sub.t].

One should, however, be aware of the simplifications and limitations of such a VAR technique. In particular, the identified nominal and real shocks do not necessarily have a direct relationship to aggregate demand and supply disturbances.

4. Results

This section starts with an illustration of the dynamics of real wages m response to real (permanent) and nominal (transitory) shocks, followed by our estimate of real wage flexibility--calculated by variance decomposition. The last part of the section assesses stability of the results over time.

[FIGURE 2 OMITTED]

4. I. Impulse responses of real wages

Using the parameters of equations (8) and (9) estimated for each of the 24 countries in our sample for the VAR decomposition described above, figure 2 shows the expected reactions of real wages in each country to one standard deviation innovations in real (permanent)and nominal (transitory) shocks, [[epsilon].sup.R.sub.t] and [[epsilon].sup.N.sub.t] respectively," ' over the forecast horizon from one to sixteen quarters. In order to facilitate a cross-country comparison, impulse response functions (IRFs) are plotted on the scale from -1 per cent to 2 per cent for the EA-12 and from -1 per cent to 6 per cent for the NMS-12. First, the stability of the estimated VARs is confirmed by the fact that all IRFs converge to some constant level. (9) Yet the speed of convergence to the constant level varies from country to country, as well as the magnitude of those constants. The long-term IRFs of real wages to real shocks range from 1 to 6 per cent. In general the effects of shocks on real wages are more substantial in the NMS-12, largely because one standard deviation innovation in shocks is larger in these countries, which is consistent with higher real wage growth in the NMS-12 compared to the EA-12 (see e.g. figure 1).

Even though the specification of the SVAR model does not impose any restriction on the sign of the impulse responses, real wages react positively to a positive real (permanent) shock in all countries, the same result as reported in Moore and Pentecost (2006).

[FIGURE 3 OMITTED]

However, the cyclical behaviour of real wages becomes more heterogeneous when the sample is extended to 24 EU countries. By construction, the response of real wages to nominal shocks dies out over time. The short-run effect of nominal shocks on real wages illustrates the relative price/ wage stickiness. The development of real wages in response to a nominal shock ('cyclicality of real wages') is crucially affected by the degree of relative price and nominal wage stickiness. In reality, both sticky wages and sticky prices exist hand in hand. Thus, the final impact of a demand shock on the economy is critically affected by the degree of price and wage rigidities.

Economic theory says that the reaction of the real wage rate to a nominal shock could be positive, negative or close to zero. While, in a sample of eight EU countries, Moore and Pentecost (2006) find just one example of a negative reaction of real wages to a nominal shock (the case of Slovakia), our results confirm the results for Slovakia and, in addition, show that this is not the only case when one considers a larger set of countries. Figure 2 shows that in the short run the IRFs of real wages to nominal shocks range between--l.5 to 4 per cent. There are six countries of the NMS-12 group (Bulgaria, Estonia, Hungary, Latvia, Poland and Slovakia) and five representatives of the EU- 12 (Belgium, Germany, Ireland, Italy and Spain) in which real wages drop down after a nominal shock. In these countries nominal wages seem to be stickier compared with prices. (10) Next, in three new member states (the Czech Republic, Lithuania and Romania) and in Luxembourg, real wages tend to rise in reaction to a nominal shock. For these countries, wages seem to be more flexible than prices. In the remaining three new member states (Cyprus, Malta and Slovenia) and six Euro Area members (Austria, Finland, France, Greece, Netherlands and Portugal) real wages do not react to nominal shocks or it is difficult to decide in which direction real wages develop after a nominal shock.

[FIGURE 4 OMITTED]

Thus, the reaction of real wages to a nominal shock appears to be rather heterogeneous. Differences in the results could be driven by specific labour market conditions and different wage formation processes across the EU countries. Indeed, the flexibility of wage-setting mechanisms could be affected both by common macroeconomic shocks and country-specific developments which are not explicitly taken into account in our analysis. (11) We find evidence of both sticky prices and sticky wages. In some countries nominal wage rigidities prevail, in the others prices are stickier than nominal wages. The overall transitory dynamics of real wages in the EA-12 seems to be smoother compared to the NMS-12. But while for the 'core' EA-12 countries (e.g. Germany, France and Belgium) the IRFs are smooth, the adjustment of peripheral countries (for example Greece, Spain and Finland) is longer and/or more volatile. However, there are some countries within the group of NMS-12 (particularly the Czech Republic, Hungary and Estonia) whose responses to shocks are smoother and faster compared to the peripheral EA-12 countries.

4.2. Real wage flexibility- variance decomposition

While impulse responses allow an illustration of the dynamic effects of shocks on real wages, variance decomposition measures the relative contribution of real and nominal shocks to fluctuations in real wages. Real wages are said to be flexible if their variation is mainly due to real shocks. Figure 3 shows the percentage of forecast variance in real wages explained by real (permanent) and nominal (transitory) shocks, at the horizon from one to sixteen quarters. At each horizon, the contribution of nominal and real shocks to the variance of real wages sums to 100. Several observations follow from figure 3. First, both groups--the NMS-12 and the EA-12--are characterised by a variety of outcomes. For the NMS-12, the variance decomposition of real wages shows that the percentage of variance explained by real shocks varies from about 40-50 per cent (Lithuania, Romania) to more than 90 per cent (Bulgaria, Estonia, Slovakia, Slovenia). In the EA-12, real wage flexibility is higher in the 'core' countries such as Germany, France and Denmark (over 90 per cent) and somewhat lower for 'peripheral' countries such as Greece and Spain (65 per cent--70 per cent after four years).

Secondly, the contribution of shocks to the variance of real wages may depend on the forecast horizon. For example, fluctuations in real wages are almost entirely due to real shocks one quarter ahead for Cyprus and Malta, but the impact of real shocks on real wage variance drops to 80 per cent and 65 per cent respectively at the four-year horizon. Such an outcome corresponds to a delayed path-through of nominal shocks into real variables. On the other hand, in for example Bulgaria, Estonia, Slovakia and Slovenia, the contribution of real shocks to real wage variance--real wage flexibility--remains at nearly constant levels, close to 100 per cent, over all time horizons.

On average, in terms of variance decomposition at the horizons up to four years, the NMS-12 are positioned between the 'core' and 'periphery'. Excluding Romania and Lithuania, the majority of the NMS-12 are characterised by higher real wage flexibility than e.g. in Greece and Spain, but lower than in Germany and France.

4.3. Changing real wage flexibility

The analysis of Moore and Pentecost (2006) ends in 2003/2004, which is just before the big wave of EU enlargement. Extension of the sample up to 2007 allows us to check the robustness of the results over time. Structural changes, such as in particular the EU enlargement, could affect the degree of wage flexibility (see e.g. Pentecost and Sessions, 2002).

We assess the stability of the results over time by comparing the estimates for the full sample (1996Q1-2007Q3) and two periods, which correspond to the pre-EU enlargement (1996Q1-2004Q2) and the euro phase (1999Q1-2007Q3). For each of the two shorter periods we find stable VARs, similar impulse response functions and calculate the variance decomposition (available upon request). Differences in wage flexibility (variance decomposition) between the later and earlier periods are illustrated in figure 4.

In general, the NMS-12 experienced an increase in real wage flexibility. Positive changes to wage flexibility in e.g. the Czech Republic, Lithuania, Malta and Poland are more pronounced than real wage flexibility declines in the cases of Cyprus, Romania and Slovenia. Oil the other hand, the EA-12 group is characterised predominantly by declines in real wage flexibility, particularly noticeable for Finland, Germany, Italy and Spain. Overall, wage flexibility in the Euro Area has diminished.

One possible explanation for this decrease in wage flexibility could be the limited pace of labour market reforms, as compared to progress in product markets deregulation (OECD, 2004). Evidence of the limited ability of real wages to adjust to shocks and the strong heterogeneity of wage adjustment patterns across the EU is reported by studies employing alternative measures of wage flexibility, based upon reaction of real wages to macroeconomic variables. For example, Arpaia and Pichelmann (2007) estimate a Phillips-curve type wage equation across the Euro Area countries and find insufficient degree of wage flexibility in the Euro Area. In addition, their results support our findings of a significant degree of cross-country heterogeneity across Euro Area countries.

Another factor contributing to a decline in wage flexibility could be related to recently rising heterogeneity in inflation rates across the EU, documented in Bulir and Hurnik (2008). Inflation acceleration, which occurs in a number of EU economies, creates cost-push pressures and leads to monetary transmission inefficiencies. Heterogeneity in inflation rates, in turn, transmits into stronger demand shocks, in particular for the Euro Area countries which share common monetary and exchange rate policies. In the presence of nominal inertia (price/wage stickiness), nominal shocks could have effects on real variables including real wages, which correspond to the observed decrease in real wage flexibility.

5. Conclusions

In this paper we have measured aggregate real wage flexibility in the NMS-12 as compared to the EA-12, following the structural VAR decomposition used by Moore and Pentecost (2006). Real wage flexibility is defined as the percentage of variance in real wages explained by real (permanent) shocks. We find strong heterogeneity of wage adjustment and wage flexibility within the country groups. Overall, real wage flexibility in the new EU members is somewhat lower compared to the core Euro Area countries, but is higher than in some of the peripheral members of the Euro Area.

Based on real wage flexibility alone, one cannot unambiguously asses the suitability of new EU member states for euro adoption. Concerning the three of the NMS12, which recently adopted the euro (Slovenia since 2007, Cyprus and Malta since 2008), only Slovenia had real wage flexibility comparable to the level of the 'core'; the degree of real wage flexibility m Cyprus and Malta is similar to that in Greece and Spain. The results of our study also suggest that joining the Euro Area is not likely to lead automatically to higher real wage flexibility.

REFERENCES

Arpaia, A. and Pichelmann, K. (2007), 'Nominal and real wage flexibility in EMU', International Economics and Economic Policy, 4, 3, pp. 299-328.

Babetskii, I., Boone, L. and Maurel, M. (2004), 'Exchange rate regimes and supply shock asymmetry: the case of the accession countries', Journal of Comparative Economics, 32, 2, pp. 212-29.

Bayoumi, T. and Eichengreen, B. (1996), 'Operationalizing the theory of optimum currency areas', CEPR Discussion Paper, No. 1484.

Blanchard, O.J. and Quah, D. (1989), 'The dynamic effects of aggregate demand and supply disturbances', American Economic Review, 79(4), pp. 655-73.

Blanchflower, D.G. (2001), 'Unemployment, well-being and wage curves in Eastern Europe', Journal of the Japanese and International Economies, 15, 4, pp. 364-402.

Boeri, T. and Brucker, H. (2005), 'Why are Europeans so tough on migrants?' Economic Policy, 44, pp. 630-703.

Boeri, T., Burda, M. and Kollo, J. (1998), Mediating the Transition: Labour Markets in Central and Eastern Europe, London, Centre for Economic Policy Research.

Bulir, A. and Hurnik, J. (2008), 'Why has inflation in the European Union stopped converging?', Journal of Policy Modeling, 30, pp. 341-47.

Dessy, O. (2005), 'Nominal wage rigidity in Europe: estimates and institutional cause', Departmental Working Papers 2005-09, Department of Economics, University of Milan, Italy.

Dickens, W.T., Gotte, L., Groshen, L., Holden, S., Messina, J. and Mark E. (2006), 'How wages change--micro evidence from the International Wage Flexibility Project', Working Paper Series 697. European Central Bank.

Enders, W. (2004), Applied Econometric Time Series, Second Edition, New York, John Wiley & Sons.

European Commission (2003), 'Wage flexibility and wage interdependence in EMU: some lessons from the early years', in The EU Economy: 2003 Review, pp. 153-200, DG ECFIN/391/03-EN, Brussels, November.

Fidrmuc, J. (2004), 'Migration and regional adjustment to asymmetric shocks in transition economies', Journal of Comparative Economics, 32, pp. 230-247.

Flek, V. (ed.) (2004), Anatomy of the Czech Labour Market: From Over-employment to Under-employment in Ten Years? Czech National Bank, Prague.

Groshen, E.L. and Schweitzer M.E. (1996), 'Macro- and microeconomic consequences of wage rigidity', Federal Reserve Bank of Cleveland Working Paper, No. 9607.

Guide, A.M., Kahkonen, J., and Keller, P.M. (2000), 'Pros and cons of currency board arrangements in the lead-up to EU accession and participation in the Euro Zone', IMF Policy Discussion Papers, No. I.

Hallett, H.A.J., Ma, Y., and Demertzis, M. (2000), 'The single currency and labour market flexibility: a necessary partnership?' Scottish Journal of Political Economy, 47, 2, pp. 141-55.

HM Treasury (2003), 'EMU and labour market flexibility' in UK Membership of the Single Currency: FJVIU Studies (June), http:/ /www.hmtreasury.gov.uk/

Hyclak, T. and Johnes, G. (1992), 'Wage flexibility and unemployment dynamics in regional labor markets', W.E Upjohn Institute for Employment Research, Kalamazoo, Michigan.

Kandil, M. (1996), 'Sticky wage or sticky price? Analysis of the cyclical behavior of the real wage', Southern Economic Journal, 66, 2, pp. 197-212.

Kopits, G. (1999), 'Implications of EMU for exchange rate policy in Central and Eastern Europe', IMF Working Paper, No. 9.

Mankiw, N.G., and Romer, D. (1991), New Keynesian Economics, 2 vols, MIT Press.

Moore, T. and Pentecost, E.J. (2006), 'An investigation into the sources of fluctuation in real and nominal wage rates in eight EU countries: a structural VAR approach', Journal of Comparative Economics, 34, 2, pp. 357-76.

Obstfeld, M. (1997), 'Europe's gamble', Broakings Papers on Economic Activity, 2, pp. 241-300.

OECD (2004), OECD Employment Outlook 2004, Paris, Organisation for Economic Cooperation and Development.

Pentecost, E.J. and Sessions, J.G. (2002), 'Changing labour market flexibility in the European Union', Weltwirtschaftliches Archiv--Review of World Economics, 138, I, pp. 148-60.

Pissarides, C. A. (1997), 'The need for labor-market flexibility in a European Economic and Monetary Union', Swedish Economic Policy Review, 4, 2, pp. 513-45.

NOTES

(1) See Fidrmuc (2004) for evidence on labour migration in the Czech Republic, Hungary, Poland, and Slovakia, in comparison with Italy, Spain, and Portugal. A detailed analysis of the Czech case is available in Flek (2004). The reasons for the restrictions on migration within the EU are discussed in Boeri and Brucker (2005); one explanation is that when the labour market is rigid, immigration may increase unemployment among the native population.

(2) Due to a lack of comparable data across countries, this paper does not attempt to perform econometric estimates of wage flexibility at the micro level.

(3) In the SVAR literature, the terms 'real' and 'permanent' shocks are often used interchangeably. The same goes for 'nominal' and 'transitory' shocks. In fact, shocks are defined according to their effect (which could be permanent or transitory) on the variable of interest, real wage in our case. We will discuss this in detail in Section 3.

(4) Among the NMS- 12 group, Slovenia adopted the euro in January 2007, followed by Cyprus and Malta in January 2008. Estonia, Latvia, Lithuania and Slovakia are 'in waiting' for euro adoption and currently participating in the new exchange rate mechanism (ERM2).

(5) A popular description of the identification strategy is provided, for example, in Enders (2004). Due to space limitations, the results of the unit root and stationarity tests are not reported here, but are available upon request.

(6) See, e.g. Babetskii, Boone and Maurel (2004) for an assessment of supply and demand shock asymmetry in the EU accession countries.

(7) Mankiw and Romer (1991) provide further readings on New Keynesian economics.

(8) We select K according to the Akaike and Schwarz information criteria, which suggest two, in some cases three or four lags. Then, we check the VAR for stability (characteristic units should lie within the unit circle) and perform diagnostic checking of the residuals for higher-order serial correlation (Ljung-Box test) and normality (Jarque-Bera test).

(9) The results of the formal VAR stability tests are available upon request.

(10) Regarding Poland and Italy, Moore and Pentecost (2006) find a positive reaction of real wages to nominal shocks. Our finding of a negative response of real wages to nominal shocks is largely due to a difference between total labour cost and wage measures. For example, when we used average monthly wages instead of labour costs for Poland and Italy, we also obtained procyclical real wages.

(11) Investigation of the determinants of wage flexibility could be an attractive avenue for future research.

Jan Babecky * and Kamil Dybczak *

* Czech National Bank, e-mail: jan.babecky@cnb.cz. ** European Commission, e-mail: kamil.dybczak@ec.europa.eu. This paper represents the authors' own views and should not be construed as representing those of the Czech National Bank or the European Commission. The authors are grateful to Giuseppe Bertola, Alex Cukierman, Kamil Galuscak, Dawn Holland, Theodora Kosma, Ana Lamo, Frank Smets and seminar participants at the ECB Wage Dynamics Network for discussion and helpful comments. However, all errors and omissions remain entirely the responsibility of the authors.